readcluster.cpp

   1 /*
   2  *  readcluster.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/28/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "readcluster.h"
  11
  12 /***********************************************************************/
  13
  14 ReadCluster::ReadCluster(string distfile, float c, string o, bool s){
  15                 m = MothurOut::getInstance();
  16         distFile = distfile;
  17                 cutoff = c;
  18                 outputDir = o;
  19                 sortWanted = s;
  20                 list = NULL;
  21 }
  22
  23 /***********************************************************************/
  24
  25 int ReadCluster::read(NameAssignment*& nameMap){
  26         try {
  27
  28                 if (format == "phylip") { convertPhylip2Column(nameMap); }
  29                 else { list = new ListVector(nameMap->getListVector());  }
  30
  31                 if (m->control_pressed) { return 0; }
  32
  33                 if (sortWanted) {  OutPutFile = m->sortFile(distFile, outputDir);  }
  34                 else {  OutPutFile = distFile;   } //for use by clusters splitMatrix to convert a phylip matrix to column
  35
  36                 return 0;
  37
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "ReadCluster", "read");
  41                 exit(1);
  42         }
  43 }
  44 /***********************************************************************/
  45
  46 int ReadCluster::convertPhylip2Column(NameAssignment*& nameMap){
  47         try {
  48                 //convert phylip file to column file
  49                 map<int, string> rowToName;
  50                 map<int, string>::iterator it;
  51
  52                 ifstream in;
  53                 ofstream out;
  54                 string tempFile = distFile + ".column.temp";
  55
  56                 m->openInputFile(distFile, in);  m->gobble(in);
  57                 m->openOutputFile(tempFile, out);
  58
  59                 float distance;
  60                 int square, nseqs;
  61                 string name;
  62                 vector<string> matrixNames;
  63
  64                 string numTest;
  65                 in >> numTest >> name;
  66
  67                 if (!m->isContainingOnlyDigits(numTest)) { m->mothurOut("[ERROR]: expected a number and got " + numTest + ", quitting."); m->mothurOutEndLine(); exit(1); }
  68                 else { convert(numTest, nseqs); }
  69
  70                 rowToName[0] = name;
  71                 matrixNames.push_back(name);
  72
  73                 if(nameMap == NULL){
  74                         list = new ListVector(nseqs);
  75                         list->set(0, name);
  76                 }
  77                 else{
  78                         list = new ListVector(nameMap->getListVector());
  79                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  80                 }
  81
  82                 char d;
  83                 while((d=in.get()) != EOF){
  84
  85                         if(isalnum(d)){
  86                                 square = 1;
  87                                 in.putback(d);
  88                                 for(int i=0;i<nseqs;i++){
  89                                         in >> distance;
  90                                 }
  91                                 break;
  92                         }
  93                         if(d == '\n'){
  94                                 square = 0;
  95                                 break;
  96                         }
  97                 }
  98
  99                 if(square == 0){
 100
 101                         for(int i=1;i<nseqs;i++){
 102                                 in >> name;
 103                                 rowToName[i] = name;
 104                                 matrixNames.push_back(name);
 105
 106                                 //there's A LOT of repeated code throughout this method...
 107                                 if(nameMap == NULL){
 108                                         list->set(i, name);
 109
 110                                         for(int j=0;j<i;j++){
 111
 112                                                 if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
 113
 114                                                 in >> distance;
 115
 116                                                 if (distance == -1) { distance = 1000000; }
 117
 118                                                 if(distance < cutoff){
 119                                                         out << i << '\t' << j << '\t' << distance << endl;
 120                                                 }
 121                                         }
 122
 123                                 }
 124                                 else{
 125                                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 126
 127                                         for(int j=0;j<i;j++){
 128
 129                                                 if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
 130
 131                                                 in >> distance;
 132
 133                                                 if (distance == -1) { distance = 1000000; }
 134
 135                                                 if(distance < cutoff){
 136                                                         out << i << '\t' << j << '\t' << distance << endl;
 137                                                 }
 138                                         }
 139                                 }
 140                         }
 141                 }
 142                 else{
 143                         for(int i=1;i<nseqs;i++){
 144                                 in >> name;
 145                                 rowToName[i] = name;
 146                                 matrixNames.push_back(name);
 147
 148                                 if(nameMap == NULL){
 149                                         list->set(i, name);
 150                                         for(int j=0;j<nseqs;j++){
 151                                                 if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
 152
 153                                                 in >> distance;
 154
 155                                                 if (distance == -1) { distance = 1000000; }
 156
 157                                                 if(distance < cutoff && j < i){
 158                                                         out << i << '\t' << j << '\t' << distance << endl;
 159                                                 }
 160                                         }
 161                                 }
 162                                 else{
 163                                         if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 164
 165                                         for(int j=0;j<nseqs;j++){
 166                                                 if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
 167
 168                                                 in >> distance;
 169
 170                                                 if (distance == -1) { distance = 1000000; }
 171
 172                                                 if(distance < cutoff && j < i){
 173                                                         out << i << '\t' << j << '\t' << distance << endl;
 174                                                 }
 175
 176                                         }
 177                                 }
 178                         }
 179                 }
 180
 181                 list->setLabel("0");
 182                 in.close();
 183                 out.close();
 184
 185                 if(nameMap == NULL){
 186                         nameMap = new NameAssignment();
 187                         for(int i=0;i<matrixNames.size();i++){
 188                                 nameMap->push_back(matrixNames[i]);
 189                         }
 190                 }
 191
 192
 193                 ifstream in2;
 194                 ofstream out2;
 195
 196                 string outputFile = m->getRootName(distFile) + "column.dist";
 197                 m->openInputFile(tempFile, in2);
 198                 m->openOutputFile(outputFile, out2);
 199
 200                 int first, second;
 201                 float dist;
 202
 203                 while (in2) {
 204                         if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(tempFile); m->mothurRemove(outputFile); return 0; }
 205
 206                         in2 >> first >> second >> dist;
 207                         out2 << rowToName[first] << '\t' << rowToName[second] << '\t' << dist << endl;
 208                         m->gobble(in2);
 209                 }
 210                 in2.close();
 211                 out2.close();
 212
 213                 m->mothurRemove(tempFile);
 214                 distFile = outputFile;
 215
 216                 if (m->control_pressed) {  m->mothurRemove(outputFile);  }
 217
 218                 return 0;
 219         }
 220         catch(exception& e) {
 221                 m->errorOut(e, "ReadCluster", "convertPhylip2Column");
 222                 exit(1);
 223         }
 224 }
 225 /***********************************************************************/
 226
 227 ReadCluster::~ReadCluster(){}
 228 /***********************************************************************/
 229